Leveraging Effective Query Modeling Techniques for Speech Recognition and Summarization

Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 1474–1480, October 25-29, 2014, Doha, Qatar. c⃝2014 Association for Computational Linguistics Leveraging Effective Query Modeling Techniques for Speech Recognition and Summarization Kuan-Yu Chen*†, Shih-Hung Liu*, Berlin Chen#, Ea-Ee Jan+, Hsin-Min Wang*, Wen-Lian Hsu*, and Hsin-Hsi Chen† *Institute of Information Science, Academia Sinica, Taiwan †National Taiwan University, Taiwan #National Taiwan Normal University, Taiwan +IBM Thomas J. Watson Research Center, USA {kychen, journey, whm, hsu}@iis.sinica.edu.tw, berlin@ntnu.edu.tw, hhchen@csie.ntu.edu.tw, ejan@us.ibm.com Abstract Statistical language modeling (LM) that purports to quantify the acceptability of a given piece of text has long been an in- teresting yet challenging research area. In particular, language modeling for infor- mation retrieval (IR) has enjoyed re- markable empirical success; one emerg- ing stream of the LM approach for IR is to employ the pseudo-relevance feedback process to enhance the representation of an input query so as to improve retrieval effectiveness. This paper presents a con- tinuation of such a general line of re- search and the main contribution is three- fold. First, we propose a principled framework which can unify the relation- ships among several widely-used query modeling formulations. Second, on top of the successfully developed framework, we propose an extended query modeling formulation by incorporating critical que- ry-specific information cues to guide the model estimation. Third, we further adopt and formalize such a framework to the speech recognition and summarization tasks. A series of empirical experiments reveal the feasibility of such an LM framework and the performance merits of the deduced models on these two tasks. 1 Introduction Along with the rapidly growing popularity of the Internet and the ubiquity of social web commu- nications, tremendous volumes of multimedia contents, such as broadcast radio and television programs, digital libraries and so on, are made available to the public. Research on multimedia content understanding and organization has wit- nessed a booming interest over the past decade. By virtue of the developed techniques, a variety of functionalities were created to help distill im- portant content from multimedia collections, or provide locations of important speech segments in a video accompanied with their corresponding transcripts, for users to listen to or to digest. Sta- tistical language modeling (LM) (Jelinek, 1999; Jurafsky and Martin, 2008; Zhai, 2008), which manages to quantify the acceptability of a given word sequence in a natural language or capture the statistical characteristics of a given piece of text, has been proved to offer both efficient and effective modeling abilities in many practical applications of natural language processing and speech recognition (Ponte and Croft, 1998; Jelin- ek, 1999; Huang, et al., 2001; Zhai and Lafferty, 2001a; Jurafsky and Martin, 2008; Furui et al., 2012; Liu and Hakkani-Tur, 2011). The LM approach was first introduced for the information retrieval (IR) problems in the late 1990s, indicating very good potential, and was subsequently extended in a wide array of follow- up studies. One typical realization of the LM ap- proach for IR is to access the degree of relevance …

Content truncated to 500 words